Abstract
The majority of visual search studies have focused on finding a target that is dissimilar from the background (e.g., finding a white T on a black background). Here we investigate the ability of participants to find a target that is similar to the background: search for a camouflaged target. Participants searched for a target (a particular toy among distractor toys) either on a blank background, or a background consisting of tiled sections of the target. Feedback was provided and participants were encouraged to decrease RTs while maintaining a high degree of accuracy. As expected, the camouflage search task proved to be extremely difficult, with RTs in the range of several seconds. However, RTs for camouflaged targets decreased and accuracy increased with training (much more so than for non-camouflaged targets). Given the high level of featural similarity between the target and the background, the optimal strategy would involve searching the background rather than distractors objects. However, in the camouflage condition participants fixated objects at rates similar to those in the blank background condition. We interpret this as an object bias; participants preferred to fixate salient objects when searching even when those objects shared fewer visual features with the target than the background. Importantly, with training this object bias decreased, likely resulting in the improvements in RT and accuracy. After training, participants were transferred to a camouflage search task in which both the target object and the camouflage background were novel. No costs were observed when participants trained to find a camouflaged target were asked to search a novel camouflage scene, suggesting transfer of training. In contrast, search times increased sharply when participants who were trained to search for non-camouflaged targets were asked to search for a camouflaged target. Practical and theoretical implications are discussed.